The conventional behavior recognition strategy for wearable sensors used in high-temperature environments typically requires an external power supply, and the manufacturing process is cumbersome. Herein, we present a rational design strategy based on fully flexible printable materials and a customized device-manufacturing process for skin-conformable triboelectric nanogenerator sensors. In detail, using high temperature-resistant ink and 3D printing technology to manufacture a coaxial triboelectric nanogenerator (C-TENG) sensor, the C-TENG exhibits high stretchability (>400%), a wide working range (>250 °C), and high output voltage (>100 V). The C-TENG can be worn on various parts of the human body, providing a robust skin–device interface that recognizes diverse human behaviors. Using machine learning algorithms, behaviors such as walking, running, sitting, squatting, climbing stairs, and falling can be identified, achieving 100% behavior recognition accuracy through the selective input and optimization of an appropriate dataset. This paper provides a research perspective for the customization, extension, and rapid fabrication of heat-resistant, fully flexible TENGs.